EURASIP Journal on Information Security

Table 2 Bona fide vs. synthetic accuracy on dataset \({\mathcal {D}_{\text {dev}}}\) for each synthetic speech algorithm

From: Synthetic speech detection through short-term and long-term prediction traces

	Bicoherence			STLT	STLT + Bicoherence
	512	256	128		512	256	128
A01	0.615	0.526	0.570	0.929	0.917	0.919	0.941
A02	0.881	0.873	0.863	0.940	0.940	0.939	0.946
A03	0.859	0.846	0.847	0.952	0.948	0.950	0.962
A04	0.546	0.505	0.499	0.886	0.827	0.879	0.915
A05	0.805	0.801	0.778	0.946	0.943	0.945	0.955
A06	0.655	0.628	0.609	0.898	0.868	0.898	0.932
All	0.726	0.695	0.687	0.926	0.907	0.921	0.942

Back to article page